Empirical research suggests that schools have significant effects on student substance use. Developed in the U.K. by Markham and Aveyard, the Theory of Health Promoting Schools (HPS) proposes processes through which school contextual and individual factors interact to affect substance use. Two constructs related to the theory, school value-added status and school ethos, have been used in recent European studies to examine school-level influences on substance use. The proposed study will apply the Theory of HPS to a U.S. context through three specific aims: 1. To establish construct validity of two measures of school context suggested by the Theory of HPS hypothesized to influence development of adolescent substance use: "school-value-added," derived from measures of student academic achievement and truancy, and "school ethos," derived from student measures of the school climate. 2. To conduct cross-sectional school-level analyses to (a) examine the extent to which there is variation between schools in student-reported substance use (i.e., alcohol, tobacco, and marijuana use assessed separately), and (b) determine the extent to which the school contextual variables, school value-added and school ethos, are related to school levels of substance use when assessed separately and together, controlling for other school-level variables (e.g. school size, school level, school type). 3. To conduct longitudinal multilevel analyses to test (a) the independent effects of school value-added status and school ethos on individual substance use, controlling for the effects of the student-level analogs of these school contextual measures-- school performance and perceived school climate --, as well as prior substance use, demographic, parental, and school compositional measures;and (b) the extent to which school-value added status and school ethos moderate the relationships between the student-level analogs of each variable and substance use. The study will use nationally representative longitudinal data from U.S. middle and high school students collected in the National Longitudinal Study of Adolescent Health (Add Health). Factor analysis, linear regression, and multilevel modeling will be used for measure development and model testing.